Comment by suddenlybananas

>That is to say, if a machine understands what you're saying and a human doesn't, that human's criticism might be below engaging with.

No wonder OpenAI accidentally made LLMs even more sycophantic if this is what people want from them.